Being Human in the Age of Intelligent Machines - Ep. 54 with Dr. Alan Lightman

[00:00] It's possible that at some time in the future that we will all have enhanced brains. So at that point, we're a new species. And, you know, it raises questions of what it means to be human. What can Homo sapiens do that this Homo techno can't do? Are there experiences that we can have, and particularly emotional experiences, that a computer can't have? You know, let's talk about love. If you've read all of the novels and all the stories, the romantic stories that have ever [00:30] you understand more about love than actually experiencing it. You can learn a lot. Language models know a lot about the world. That's beyond just what is explicitly in a particular sentence, just by the way letters combine. So I do think if it has read all the love stories, it actually knows something and something that not the same, but more than you might expect. Yeah, it might be a good relationship counselor. Yeah, I think it is, actually. I mean, I use it all the time for that. It's helpful. Yeah, because it knows about all the romances and relationships that have gone south. [01:00] yeah [01:17] Dr. Leitman, welcome to the show. Thanks for having me on, Dan. Thanks for being here. So for people who don't know you, you are a physicist and a writer. You're one of the first people at MIT to hold a joint faculty position in both science and humanities, which I think is amazing. You're the author of Einstein's Dreams, which I think is an incredible book, and most recently, the book Miraculous and the Material. So thanks for coming on.

[01:41] Thanks for having me. The thing I like about the miraculous and the material is it, it reads to me almost like a, like a devotional, but for people who are, who feel awe from science. So each chapter is like a little story about. [02:00] you know, atmosphere or atoms or bubbles. And it's, it's an alphabetical order order. And it almost feels like something I could read every day just to kind of like think a little bit about how the world works. And then, um, [02:13] Feel a little bit of awe or wonder. Tell me about that. Well, I want to make a very slight editorial correction. The title of the book is The Miraculous from the Material. Ah, did I? What did I say? The Miraculous and the Material. Sorry. Yes. The Miraculous from the Material. I apologize. An important distinction. No apologies needed. [02:32] That's an important distinction because the [02:36] The point of the book is, [02:38] Yes, that... [02:39] even though we may understand the material basis for a lot of extraordinary natural phenomena. [02:46] like spider webs and volcanoes and so on, that that doesn't make them any less awe-inspiring. [02:54] It feels like a little bit of a counterpoint to that Walt Whitman poem, The Learned Astronomer. Are you familiar with that poem? [03:01] Yes. Yeah. How do you respond to Whitman? Like if you if he was here, what would you say to him? [03:05] Well, yes, I do know about... [03:07] that poem. I love that poem. Um, [03:11] It's about, for your listeners who don't know it, it's about a

[03:15] a person listens to a lecture, [03:19] on astronomy. [03:21] and goes out on a, then on a dark night and looks at the sky and is, is just, [03:26] overwhelmed by the beauty of the sky. And [03:30] The message of that poem, I think, [03:33] is that [03:35] that explanations of phenomena don't really replace the actual experience of the phenomena. And sometimes take away from it, I think. And sometimes take away from it. There's like... [03:46] drain it of its enchantment right yeah there's a slight negativity there and i don't have that point of view [03:54] I think when you have a scientific explanation of a phenomena, it actually enhances your [04:02] appreciation. [04:03] of the phenomenon. And for me, having an explanation of spider webs or volcanoes or lightning [04:12] where the rings of Saturn doesn't diminish one iota my... [04:17] Awe and appreciation. [04:20] and admiration [04:22] of those phenomena. [04:24] I think one of the ways to restate that, which I'm curious about how you feel, is since the Enlightenment, [04:34] As we have more explanations for things, there's this common refrain that we've disenchanted the world. And when we look at the sun or whatever, you can just say, it's just...

[04:46] a collection of gases. It's not a god, right? And that just, it sort of like drains, when you have the explanation, it sort of drains everything else out of it. And the explanation is just this sort of like inert dead thing that people are like, well, that's all it is. So everything is meaningless. So how do you reframe an explanation in your head? And does it have just the just or is there something else? Well, I would just omit the word just. That would be my solution to that dilemma. [05:15] Um, [05:16] Yes, the sun is a collection of atoms and molecules and... [05:21] It balances gravitational forces inward against thermal pressures outward and all of that. [05:28] Thank you. [05:29] But isn't it amazing that, [05:32] First of all, the sun and all stars are amazing spectacles. [05:39] But isn't it amazing that human beings can understand so much about the cosmos that's far beyond science? [05:48] Our small planet. [05:50] And, you know, we're never going to be able to do experiments directly with the sun. [05:55] or with other stars, but just on the basis of our... [06:00] brains and intellectual power, we have [06:05] learn how those things work. We've learned that the universe began 14 billion years ago, which is many, many, many, many, many human lifetimes. [06:14] We've learned about the vast extent

[06:18] of the universe, which is far beyond our [06:21] tiny dot of a planet, [06:24] We've learned... [06:27] the molecule that encodes the instructions for making more human beings. And to me, all of this [06:36] All these things we've learned, all of this knowledge is a testament to the power [06:41] of the human brain, [06:43] Um, [06:45] It doesn't diminish any of the phenomena in nature that, [06:50] The fact that we understand many of those phenomena. One of the things I think that comes out of what you just said, like especially going back to omitting the just, is explanations can exist alongside of things. [07:06] our own conscious experience of things. And you don't have to like get rid of your conscious experience because you have the explanation. And they sort of, they might be mutually enriching. How do you think about that? The way that like... [07:20] Viewing reality from those different lenses [07:22] plays together or should play together? Well, I think they are mutually enriching. That is the direct experience [07:30] of the world and [07:32] a scientific explanation of it. [07:36] The scientific explanation does not replace the direct explanation. [07:40] experience. [07:42] And [07:44] Thank you. [07:46] That's why I call myself a spiritual materialist.

[07:49] The materialist part of that is my scientific side, which understands everything in terms of [07:56] atoms and molecules and the law of conservation of energy and so on. [08:01] But I'm also open to [08:03] and embrace [08:06] Spiritual experiences like feeling part of something larger than myself or the appreciation of a waterfall or a sunset or a communion with a wild animal. [08:20] or [08:21] Um, [08:23] relationships with people. [08:25] that all those things are... [08:28] vital [08:29] And part of what I call spirituality, [08:33] And I embrace all of that. [08:36] So I call myself a spiritual materialist. - I think it's a beautiful term, and there's a lot to unpack there. The first thing I wanna unpack is just what, [08:45] what you mean by spiritual. Because some of the things you listed, for example, like the beauty of a sunset or human relationships are not things that [08:52] people would typically categorize necessarily as spiritual maybe the idea of being part of something bigger than yourself maybe something that you can't explain but yeah how do you what is what is spiritual to you versus not spiritual [09:04] Well, spiritual for me is a list of experiences that I mentioned, feeling part of things larger than myself. [09:13] Uh, [09:15] The appreciation of beauty. [09:18] Um,

[09:21] Uh, [09:22] Thank you. [09:23] Thank you. [09:24] communion with God, [09:26] with animals and with non-human animals and [09:31] uh, [09:32] That all of that is the experience of awe. [09:38] That's part of my understanding of spirituality. Now, many people... [09:44] include a belief in God or some belief, [09:49] divine being as part of their understanding of spirituality. And I think that's [09:54] perfectly fine. [09:56] Uh, so, uh, I think that you can, [10:00] be a spiritual person, [10:03] whether you believe in God or not. And, you know, I'm not going to define God right now unless you ask me to. I'm about to. Have some understanding of that. So it it can either include a belief in God. [10:21] or not it can include a belief of heaven and hell or not [10:24] Uh, [10:26] But I do think that we can have my version of spirituality, whether or not [10:34] We believe in God. [10:36] Does your personal version, I'm assuming because of the word materialist in spiritual, uh, [10:44] materialism. [10:45] that you don't believe in God. [10:48] Well, I'm...

[10:51] an agnostic on that. If you draw a line with total belief and faith on one end and [11:00] agnosticism in the middle and atheism on on the other side i'm somewhere between an agnostic and an atheist [11:08] I'm not an atheist. [11:11] Um, [11:12] But I'm, [11:14] in that direction. I don't think that that our minds are [11:19] and understandings are big enough [11:22] and broad enough and deep enough to rule out the possibility of God. That is an intelligent being that created the universe. [11:31] That gets me to my next question, which is one of the interesting implications of materialism is that everything can be explained in terms of basic physical laws, right? Yeah. [11:44] Yeah. [11:45] But... [11:46] Because, for example, we don't know how consciousness arises from the activities of neurons, that's still kind of an undecided question. So declaring yourself a materialist while that's undecided requires a little bit of faith. How do you think about that? Yeah. Well... [12:03] The question of consciousness is a good one because that's one of the outstanding things. [12:08] Mysteries. [12:10] at the frontiers of science. So my view about consciousness and things like that [12:17] is that, [12:18] I believe that consciousness and, in fact, all mental experiences are

[12:23] are rooted in the material brain, that is, in the electrical and chemical activity of neurons in the brain, which are material things. [12:34] So I think that all mental sensations are rooted in that. [12:39] But we don't yet understand how you get from that material basis to the [12:48] Feeling. [12:49] that we call consciousness. Consciousness is a feeling. It's a name that we give to a certain sensation caused by the 100 billion neurons exchanging chemical and electrical signals. [13:04] It produces a certain sensation that, [13:07] And we call that consciousness. [13:10] But it's very, very hard to know and maybe impossible to know [13:15] how another organism feels. [13:18] And that, of course, is related to the question of whether AI can ever be [13:23] conscious, [13:24] And my view there, [13:27] is that any finite list of manifestations of consciousness that you write down [13:35] For example, [13:37] self-awareness, ability to plan for the future. I think that at some point that AI will check all of the boxes [13:46] of the manifestations of consciousness. But whether that computer is actually conscious is a different question. Right. I think it sort of plays with one of the themes of a lot of your work, which is there's a limit to how much we can write down or how much we can explicitly say. And then there are some things that are mysterious and maybe we can...

[14:09] feel into some of them but like even that is is is maybe is maybe too much so um you know anything that we can give a target to for ai [14:19] it can do, but there's actually probably limits to what we can give targets to. Yes, I think I would agree with that. I mean, of course, as time goes on, AI will be able to do more and more things, but [14:32] But there will always be things that it can't do. [14:36] Most likely. Yeah. I think my feeling about that, I'm sort of curious how you feel is... [14:41] At the point, like, for example, the idea that something reacts to pain, right? [14:50] is one of the ways that we might tell if something's conscious, you know, like, because of ant runs away, and when you try to kill it, we're kind of like, it probably has some amount of consciousness. And that seems like a useful thing to build into AI systems. And in fact, like, generally, like, we're doing those kinds of things with AI right now, we don't think of it as causing them pain, but we reward them for good things and punish them for bad things, that kind of thing. It seems like at some point, [15:17] Um [15:18] that sort of basic thing will ladder up into behavior that looks a lot like consciousness to us. And we may just like decide to treat it that way because we sort of. Well, I would disagree a little bit with your initial statement that the reaction of pain is. [15:36] represents consciousness because you can take an ant and,

[15:41] And I think most of us would agree that an ant doesn't have... [15:45] anything [15:46] resembling consciousness, at least our level of consciousness. [15:51] I mean, consciousness is a graded phenomenon. It's not an all or nothing thing. Dolphins can recognize themselves in the mirror. Crows play games with each other. [16:02] So there's some level of consciousness there. [16:06] But as you go down to... [16:09] the animal kingdom [16:12] you eventually get to [16:15] Ants and even single-celled organisms like amoeba, [16:19] And I would say that even in, let's take an amoeba. [16:22] That if you put... [16:24] some chemicals near an amoeba that are dangerous to it. [16:29] It will... [16:31] It will react. It will avoid those chemicals. Now, that's happening. That's a totally automatic response. That doesn't involve any higher levels of cognition. [16:44] And I think that you could do the same with a computer. Let's say that a computer needs to have a certain temperature range in order to operate. [16:56] And we know that when a room gets too hot, it's bad for a computer. [17:02] That's why we have fans on our laptops and so on. [17:07] If you had a computer sitting in a room and you turned up the temperature and made it hotter and hotter and hotter,

[17:15] Yeah. [17:16] Eventually, there would be sensors in that computer [17:19] Thank you. [17:20] that would try to turn off certain things, it would react. It would react. I mean, my my [17:28] MacBook Pro reacts when it gets hot. [17:33] So that doesn't involve any consciousness in my understanding of consciousness. Sure. I think that's totally right. Maybe there's different levels of complexity of sensitivity to pain. So... [17:48] Um, you know, and in, in the amoeba example or in the laptop example, it's about chemical, like a certain amount of chemicals or a certain temperature of the air, you know, surrounding environment. Whereas like for a human, it's like my mortgage, I might be not be able to pay my mortgage and that causes me a certain level of pain. And that represents like a, a psychological pain. Yeah. Um, so maybe something, maybe something like that, where at some point, [18:14] you know [18:15] I'm not saying that there's ever going to be a perfect test, but there might be like some of these things you can kind of like get, you get glimpses to be like, I don't know. It seems like there's something in there. Yeah. Well, I think even with physical pain, [18:27] The human reaction to physical pain is different than the reaction of an amoeba. [18:33] to, [18:34] to toxic chemicals. Because with humans, [18:38] When you... [18:39] have pain, [18:41] Okay. [18:41] You are self-aware of having pain and you're able to name it.

[18:46] You're able to name this sensation as pain. It's similar to other sensations that I've had in the past years. [18:54] that often also caused me these [18:58] reactions, these physical reactions, you can categorize it and name it. [19:05] And so your mind is operating at some higher level that is not only sensing certain things, [19:13] certain things, but it's actually aware [19:17] That it's sensing that. [19:19] and is able to name it and categorize it. So you're processing it. Yeah. There's something else happening. Yeah, there's something else happening. So that's, that's, that's a difference between a human person. [19:30] putting their hand near a fire and an amoeba [19:35] Moving away from some toxic chemicals. Yeah, yeah, yeah, yeah. That's interesting. [19:41] This leads pretty nicely into our AI discussion, which I'm professionally obligated to bring up. [19:49] So you wrote an article for The Atlantic called When the Unnatural Becomes Natural. It's about AI. And it's about the... [19:58] basically the problems and the promise of getting used to things that are artificial. Can you talk to us about that article? Yeah, I'm curious. [20:07] Curious for you to go into it with us. [20:09] Well, what we call natural and unnatural [20:13] is somewhat arbitrary.

[20:16] Thank you. [20:17] And [20:19] You could take the view... [20:22] that [20:24] Eyeglasses and hearing aids are unnatural. [20:28] because, [20:30] Thank you. [20:31] We weren't born with those abilities. [20:35] And you could take the point of view that any machine that we use [20:41] create is unnatural because that machine... [20:46] We didn't find it lying under a rock one day. [20:50] Thank you. [20:50] We put it together with intelligence. [20:55] On the other hand, and this is a point of view that I've taken our article in The Atlantic, [21:00] Thank you. [21:01] You could take the point of view... [21:03] The sense we, [21:06] are natural. We homo sapiens are natural. We evolved... [21:10] from lower organisms, [21:14] Um, [21:15] I don't think there was anything supernatural that created human beings. I think that we started off as single cell organisms in the ocean and evolved from there. [21:25] due to Darwinian evolution. [21:27] So we are totally natural. [21:30] And our brains are totally natural. [21:33] And you could take a point of view that anything that our brains invent is [21:39] is natural and [21:41] because it is... [21:42] Uh, [21:43] an inevitable consequence [21:46] of something that is natural.

[21:48] Thank you. [21:49] Um, [21:50] So that kind of blurs the distinction between, [21:53] between [21:55] human made [21:57] objects, [21:59] Yeah. [22:00] and objects that we just find on a rock. Right. And I think one of the sort of implications of that is [22:10] Maybe to say that typically when we use the word natural, we mean things that we grew up with. [22:16] Um, and you, you start to forget that, um, [22:20] books or... [22:22] you know, [22:23] keyboards or windows or whatever were at one point like technology that people might have been afraid of and probably and actually were or didn't exist at all. Exactly. Yeah. And and there's something. [22:37] There's some there's that's an interesting lens through which to view AI. So how does that help? How does that change your perspective on air? What do you bring to AI with that perspective? Well, my granddaughter can can use technology. [22:51] apps on the smartphone platform. [22:55] with more agility. [22:57] and familiarity than I can. [23:00] Thank you. [23:01] So everybody growing up in the last... [23:04] 20 years or so. [23:06] is very familiar with technology, with the internet, with smartphones, with apps on the smartphones. And that's their world.

[23:18] Um, [23:22] And of course, [23:24] Uh, [23:25] old farts like myself grew up a long time ago. So I think it's a good point that you make that it depends [23:32] somewhat on what we grew up with. Yeah. I want to take one minute away from this episode to introduce you to our sponsor, LTX Studio. I think storytelling is one of the most essential skills of the AI age. You can bring your stories to life with just a few words, complete with a cast, storyline, settings, all according to your style and your specifications. LTX Studio is helping storytellers visualize their stories in entirely new ways. Two of the most magical parts of LTX are in their character generation and storyboarding. Here's how it works. If you type in a description [24:02] and maybe add a headshot, LTX generates unique dynamic characters, each with their own distinct look and personality. Remember the old days of struggling with storyboards? LTX makes it simple. If you need to map out a bustling debate in ancient Greece or a castle under the sea, LTX lays out and expands on your vision shot by shot. Better yet, it suggests new angles and shots you might not have considered. First-person perspectives, wide angles, close-ups, you've got it all. I can switch [24:32] ultra-realistic characters or cartoon-style art, it's all just one click away. The AI revolution is just starting. But if one thing's clear, it's that it's not replacing human creativity. It's expanding it. So if you've ever had a story in your head with no way to bring it to life, [24:46] Start with LTX Studio. It might just be the creative partner you've always needed. Check out the link in the episode description for more details. Hey there, Dan here. I wanted to take a one minute break from the episode to tell you about our latest sponsor. All right, let's play a game.

[25:01] What powerhouse productivity tool is also free for individuals? [25:08] Nope, not that one. Try again. [25:14] You may not expect this, but it's Microsoft Teams. Yup, the same teams that big enterprises swear by. [25:21] also has a free plan for individuals. [25:23] Whether you're jamming on a side project or bootstrapping a startup or building a community, [25:28] Teams has all of the features that other platforms nickel and dynia for using. [25:32] You can get unlimited chat, 60-minute video meetings, file sharing, and collaborative workspaces all for free. And the real magic is that everything is integrated in one seamless collaborative workspace. That means there's no more hopping between different applications for messages, meetings, and file sharing. [25:49] Teams puts it all at your fingertips to save you time and money. So ditch the app overload and the subscription fatigue. [25:57] and use Teams to experience effortless collaboration today. [26:01] Are you ready to streamline your workflow? Head to aka.ms/every to use Teams for free. Your productivity will thank you, and so will your wallet. [26:10] And now... [26:11] Back to the episode. And just, I guess, tell us what your current experience is with AI or what your current feeling is about it. You know, you've been around a while. You've seen everything from computers to smartphones, the internet to this. You're someone that deeply understands the kind of like...

[26:32] technical mathematical foundations of the universe. Um, and all those also as a, uh, a humanist. Um, and this is sort of a new, there's a, there's something new here, at least to me, it feels like there's something new here going on. I'm curious how it feels to you. [26:47] Well, I think that [26:49] First of all, AI is developing very, very rapidly, and even the [26:54] People at the frontiers of research can't predict when certain benchmarks will be reached. [27:01] So I think most people would agree that AI has the potential for great benefits and also great dangers. [27:13] Thank you. [27:14] And, [27:15] It's going to be awfully hard to regulate. [27:19] uh [27:20] No, the European... [27:22] Union a year ago or so issued some policy for regulation. The U.S. is now investigating that [27:30] Most countries are [27:32] investigating [27:33] different kinds of regulation, [27:36] But it's going to be very, very difficult to regulate because a lot of it is being done [27:42] and private companies, [27:44] And the profit incentive is huge. [27:47] to be one step ahead in the AI game. [27:52] And, uh, [27:53] There's also national pride that some countries are going to just want to be out in front with AI. [27:59] and are going to resist [28:01] any constraints. So,

[28:05] It's going to be very hard to regulate. [28:07] And, [28:09] That is a problem. [28:11] Um, [28:15] I think having AI that can make autonomous decisions on the battlefield is very dangerous. [28:24] Uh, [28:26] Thank you. [28:27] It's possible... [28:30] that, you know, to have Darwinian evolution, you need [28:34] mutations [28:36] and you need reproduction. [28:40] AI is almost capable of that now. [28:45] Thank you. [28:45] I mean, [28:47] One AI can control a factory that's making other AIs. [28:53] So once you have those two necessary ingredients for Darwinian evolution, you [29:00] Then, you [29:02] It's automatic that the organism, in this case, the organism is made out of silicon, is [29:07] can develop a sense of self-preservation. [29:12] which also comes from Darwinian evolution. [29:15] And so at some point, [29:17] Advanced AI may decide in the interest of self-preservation that we homo sapiens are not good for it. [29:24] Thank you. [29:25] Thank you. [29:27] Makes sense. I think, I think that's a, I think that's a reasonable viewpoint. It's also, it's, it's, um, [29:32] I think one of the things that that, [29:35] perspective has not

[29:37] yet grappled with, or one of the reasons why it hasn't tried to kill us all yet, is it doesn't quite account for [29:45] the way in which that self-preservation instinct is also being created in the context of an instinct to collaborate and coordinate with humans that is deeply, deeply, deeply embedded. [30:00] um and [30:03] So I'm not saying it's sort of an either or thing, but I think sometimes those thought experiments miss that the co-evolution happening highly incentivizes these systems from the beginning to want to be helpful and around us. Well, of course, we hope that that's going to be the case. [30:21] But I agree with you. Those are two forces that will both be operating. [30:27] And of course, this ability of AI to develop a sense of self-preservation, we're not at that point yet. [30:35] But we may be at that point at some time in the future. Yeah, that makes sense. One of the things that I'm kind of curious to talk to you about is that this – [30:45] sort of AI revolution or [30:47] this new level of technology might change what it means to know things or how we might know things. So particularly in science. So, for example, right now, if you're in, you know, working in psychology or neuroscience and you're asking a question like, what?

[31:08] what is depression. You have to go find one particular mechanism for depression, which has proven to be really hard to find. [31:19] But what you can do instead... [31:22] is train a model on a bunch of people who start out without depression and then get it, and have the model be able to predict who's going to get depression and who's not going to get depression, which feels like a different way of knowing things than finding root explanations. [31:38] Have you thought about that? Like, what does that, what does that bring up for you? [31:42] Well, that's... [31:44] That's like a data lookup table. [31:48] You know, you get lots and lots of data on different people. [31:52] And you correlate that data with the data has both some information about the degree of [31:58] depression and also depression. [32:01] uh, [32:02] Lots of background data on the person where they grew up. [32:06] you know, who their parents were, et cetera, et cetera. [32:09] And of course, a computer can sort through all of that. [32:14] information much faster than a human can. [32:17] And, [32:19] I call that not thinking, but thinking. [32:21] data lookup. [32:25] Um, [32:26] Of course, it can still be very, very beneficial to us, even if we don't call it thinking. And we know that in the medical field that, [32:35] that AI is already being very useful in developing new drugs because it can

[32:42] Try out. [32:44] Lots and lots and lots of different chemical products. [32:49] combinatorics and possibilities [32:52] and find out which ones have different properties, you know, [32:55] this would be theoretical, [33:00] It's not actually doing experiments, but it can look at the different shapes of molecules, especially when you have protein folding. [33:09] It can try out lots and lots of different configurations [33:13] And, [33:15] It's just trying out lots and lots of things. [33:18] And it's a totally mindless thing. [33:22] operation, [33:23] Thank you. [33:24] But it can be extremely useful to the medical profession in finding new drugs. Let me push you a little bit on the data lookup thing. Because I think that's a common... [33:37] That's a common worry that people have about, I don't know, neural networks or the way that AI works. And to some degree, I think there's something to what you're saying, but like... [33:46] Um, [33:48] To make a data lookup table work, let's say for depression prediction, you'd have to have [33:54] a database that's so big and so hard to look through that it probably wouldn't... If it was actually data lookup, that it probably wouldn't work. And the way that a neural network works is it distributes all of that context and all of that background information across the entire network where it doesn't exist in any one particular place. And it can process in parallel lots and lots and lots of tiny correlations between different pieces of context that might lead to

[34:24] So like, [34:25] You might have a thousand different things that might lead to depression and three of them might be turned on in any given situation. [34:31] any given... [34:32] you know, [34:33] person. And that feels a little bit more like what a neural network is doing than data lookup. [34:39] Yeah, I agree with you. So I accept that that clarification and revision what I said that is more than a data lookup. [34:48] table, though when [34:51] They're different. [34:53] elements of different digital digital neurons. [34:56] of a neural network can exchange information back and forth. [35:02] in a nonlinear way. [35:05] then that is [35:06] More than just data lookup, the computer is actually teaching itself. Yeah. So then is that in your mind thinking? Because that is how like a transformer works. Well, that certainly comes closer to thinking. [35:21] I mean, I think the... [35:24] Thank you. [35:25] How you define thinking is somewhat arbitrary. [35:29] But I think that that's a lot closer to thinking than thinking. [35:34] than just a data lookup table. [35:36] Thank you. [35:36] I think when the [35:38] neural network starts modifying itself. [35:43] due to the back and forth communication of the digital neurons, that that's [35:48] Certainly approaching thinking. [35:50] It's definitely doing that in the... [35:54] when it's being trained, although it's not modifying itself, it's, it's being, you know, someone else, something else is modifying it, but the, the, the,

[36:01] union of the training program in the network is doing that. But yeah, once it's trained, it's not. [36:07] it's not being updated, at least for now. Although there are architectures that are, [36:12] kind of getting close to that. But one of the things that... [36:16] this line of thinking kind of takes me down, which I think is related to your work, which I think is a lot about how even when you're [36:24] and you tell me if I'm wrong, but even when you're dealing in the most [36:28] sort of abstract, rational places in human knowledge. So like physics, human intuition and creativity is still incredibly important. [36:39] Yes. And one of the things that this makes me think of, and when we talk about new ways of knowing, this is why I'm kind of interested in AI stuff is... [36:49] For example, the ability to predict who's going to get depression without an explanation, which is what an AI model might be able to do. [36:57] Um, and, and probably can like a neural network probably can in some, in some circumstances do it pretty well. It actually looks to me a lot like human intuition. Um, [37:05] like a skilled clinician being able to tell like, this is what's going on with someone because they've seen thousands and thousands and thousands of examples. But the difference is that intuition is stuck in, [37:15] in our heads. [37:17] And so in order to make that useful to other people, we've had to create... [37:22] theories or like, you know, mathematical explanations. And, you know, [37:27] Models seem to be an alternative way to express things that are very difficult to express in terms of mathematical theories.

[37:38] What do you think about that? [37:39] I agree with that. [37:44] that models... [37:46] of intuition or depression [37:50] Uh, [37:52] can reach conclusions. [37:55] that, [37:56] We cannot articulate [37:58] Thank you. [37:59] Um, [38:01] I mean, another way to... [38:03] to explore the origins of depression and be able to predict depression is for a neuroscientist [38:12] to see what [38:15] molecular processes, [38:18] or associated with [38:20] and brains that have depression. [38:24] And I know some neuroscientists are doing exactly that and focused on depression. [38:31] And they're looking for a chemical product. [38:36] basis. [38:38] And the brain. [38:40] that leads to depression. [38:42] So that's another point. [38:45] That's another avenue. You're not looking at [38:49] you're not doing correlations with the background of lots and lots of people and they're [38:54] behavioral characteristics, you are [38:58] looking at, you know, in this case, maybe one or two brains and trying to understand them in detail. Yeah. My guess is that... [39:07] The word depression is actually a bunch of different things. Yes, I'm sure you're right.

[39:12] And that... [39:14] There are probably cases where you'll be able to find some specific physiological marker, but it's probably going to be the combination of a physiological marker and a bunch of other context in a person's life. And it won't be just localized in that, just the chemicals. Yeah, I think you're right about that. Yeah. [39:34] Um... [39:37] So you have a documentary series called Searching, and you bring up the idea of homo techno. [39:44] Can you talk about what that is? Well, uh, well, [39:48] We've been... [39:50] Human beings or homo sapiens have bypassed Darwinian evolution now for a few hundred years or so, or maybe longer. You know, just hearing aids and eyeglasses. [40:03] or an example of [40:06] how we [40:08] We're no longer subject to... [40:11] to characteristics have survival benefit because we can create new devices [40:20] And we can develop medicines that will allow people to survive that would have died forever. [40:26] several hundred years ago. [40:28] and so we're, we're, [40:33] We're evolving by our own hand. [40:37] Um, [40:39] The AI is just one example of that.

[40:44] And I think that at some point in the future, [40:48] that we will have hybrid... [40:51] Organisms. [40:52] that I call homo techno. [40:55] Thank you. [40:55] that have evolved beyond Homo sapiens and are part human and part machine. [41:02] And just for example, [41:04] We already have the ability to implant computer chips into the brains of paralyzed people. [41:11] that allow them to [41:14] Move robotic arms simply by pure thought. [41:20] And even control the movement of the arm by pure thought. [41:25] And so... [41:27] It's possible that at some time in the future that we will all have enhanced brains and [41:33] I mean, already we have enhanced eyesight and enhanced hearing with hearing aids and eyeglasses. Which are part of your brain. Yeah. [41:41] Right. So we it sometime in the future, we may have computer chips implanted in our brains. [41:50] that, for example, allow us to be connected [41:54] instantly to the internet. [41:56] to the vast amount of information on the Internet. [42:00] or to compute, to communicate with other people, [42:06] through the internet. So I'm thinking a thought and [42:10] And I'm connected to the Internet. And I my thought is broadcast to the Internet and then broadcast to the computer chip in your brain.

[42:19] And it would be another form of communication much faster. [42:24] Thank you. [42:24] Um, [42:25] So, yeah, [42:26] Thank you. [42:27] At that point, [42:29] We're a new species. [42:31] And, you know, and... [42:34] You know, I'm just suggesting that we would be a new species then, which I call homo techno. And why is that interesting to you? Or what do you do with that, thinking about that? Well... [42:44] It raises questions of what it means to be human. [42:48] And I think that AI is doing that already. [42:52] You know, what can we do that that AI can't do? [42:56] Or what can... [42:58] Homo sapiens do that this homo techno [43:01] can't do? Are there experiences that we can have, and particularly emotional experiences, [43:09] that a computer [43:10] can't have. You know, let's talk about love. [43:14] Well, a computer could read every novel that's ever been written about love. [43:20] Right. All the love affairs throughout the last several thousand years of history. [43:26] Would it understand love itself? [43:30] the same way that you and I understand it, who have fallen in love. [43:35] and know that thrill... [43:38] That feeling... [43:41] which is very personal. I mean, each love is different. Each person falls in love in a slightly different way.

[43:51] And can that experience... [43:56] ever be [43:57] replaced by a computer. [44:00] If you've read all of the novels and all the stories, the romantic stories that have ever been written, [44:08] Do you understand more about love than, [44:11] than actually experiencing it? I think it's a great question. And I think, um, [44:17] Yeah. [44:18] My answer to that is, [44:20] is... [44:21] Well, obviously no in a certain sense because assuming these things are not conscious, then no. But... [44:28] One of the things that I think is really interesting about language models is there's, it shows that there's a lot more to the way that text is put together than what it explicitly says. You can learn a lot. [44:39] Language models know a lot about the world that's beyond just... [44:42] what is explicitly in a particular sentence, just by the way letters combine. So I do think if it has read all the love stories, it actually knows... [44:50] Something. [44:51] And something that is... [44:55] Not the same, but... [44:56] more than you might expect or more than I would have expected, which is really interesting. Yeah. It might be a good relationship counselor. Yeah. I think it is actually. I mean, I use it all the time for that. It's helpful. Yeah. Cause it, it can, it knows about all the, the, [45:14] the romances and relationships that have gone south. [45:21] Yeah. But there's still going to be things, I hope and believe, that,

[45:26] that you and I can experience with love that the computer [45:30] cannot duplicate. Well, that's the thing that you brought up that I think is an interesting one, which is raising the question of, okay, what does it mean to be human? And then the next thing that you brought up is to answer that question, what can humans do that an AI can't do? Which I think once you make that move, it brings us back to something that you said earlier, which is, um, [45:51] we were talking about AI can do anything that we can put our finger on, anything that we can define well. But once you ask that question, you're already setting yourself up for, well, there's nothing I can say because anything I can say, [46:08] eventually it's going to do, but there's a lot that you can't say. Right. And having some faith, I think that, um, [46:16] humans are this, we're, the problem with defining human nature is we're just static and it's not static. It's always changing. It's, it's high dimensional. It's very fluid. And it sort of requires, I think like a little bit of faith that there's a lot of things that we can't, we can't say, we can't put our finger on, but we know. And the moment we try to articulate it is the moment [46:37] we kind of lose it. [46:39] Yeah. [46:41] Well, that brings us back to Walt Whitman's The Astronomer. [46:45] point. [46:47] Um, [46:49] I like to hope that, [46:51] that there are things that I can do and can experience that a computer will not be able to do.

[46:59] I, for some reason, the thought that I can be totally duplicated, [47:05] by a computer. [47:07] bothers me. [47:10] And maybe I shouldn't be bothered by that. [47:14] But, yeah. [47:16] Um, [47:18] Thank you. [47:19] you [47:21] It's also related to the ego. [47:25] you know, [47:26] I'm bothered somewhat because I have an ego. [47:29] And the ego was developed by Darwinian evolution. I mean, it had survival benefit. [47:37] I don't know whether Freud ever talked about that aspect of the ego, but I think that [47:42] that most, um, [47:44] biologists would say that the development of the ego [47:48] the sense of a self and the allegiance to that self is, [47:53] had some kind of survival benefit at one point. [47:58] And so it's that... [48:02] Ego and sense of self that bothers me in thinking that I can be. [48:07] replaced by silicon. I agree. Which I think is, you know, it's healthy. I think there's definitely something healthy there. I think there's... [48:16] There's a question about... [48:18] Is that sense... [48:20] How will that sense evolve over time as you talked about getting used to things that are quote unquote unnatural? [48:26] So for example, I'm not bothered that you can also fall in love.

[48:30] I think that's great. We can connect about it. [48:32] Um, and, [48:34] But you're someone else that like... [48:38] In the same way that maybe in 20 years or 30 years, I might not be bothered that I think a machine can fall in love. Because I've created... Even though we kind of have the same brain, I've created this... [48:50] separate identity because we're separate people that I think, um, [48:55] is... [48:57] to some degree, inside of us once we get used to a certain thing. Once we've lived with it for a while, we're not as threatened by it anymore, and we realize, like, no, no, no, I'm still my unique self. Even if, you know... [49:08] Dr. Lightman, you've written way more books than me, and you've had way more experiences than me. It doesn't necessarily just... [49:14] eliminate my my value totally. [49:17] Does that make sense? [49:18] Yeah, there's the anecdote about the frog that you put into warm water and you slowly turn up the temperature. [49:26] And it never notices, there's never any definite point [49:30] But it's where it knows that it's being killed. But at some point, of course, it is killed. [49:38] And I wonder whether as our... [49:41] technology gets more and more advanced and we get more [49:44] accustomed to it and more familiar with it [49:48] whether there will ever be a sharp line where we realize that we've crossed a boundary, [49:56] Thank you. [50:00] And one thing, one way in which I think that we've crossed a boundary, but it's not a very clear boundary, is the pace of modern life.

[50:09] And the pace of modern life or the pace of life has always been regulated by the speed of communication. [50:15] and I, [50:18] The speed of communication has gotten faster and faster. Uh, [50:22] In the middle 19th century, the telegraph was the new communication device, and it could communicate three bits per second. [50:31] And then, [50:32] In the mid-1980s, when the Internet disappeared, [50:36] got roaring and [50:38] Uh, [50:40] You could communicate a thousand bits per second, and now it's billions of bits per second. [50:47] And I, [50:49] you can tell that the pace of life has increased. [50:52] Uh, [50:54] that we now [50:56] We look at our smartphones every five or 10 minutes. We rush around from one appointment to the next. [51:02] We rarely take the time to [51:06] to go out and take a quiet walk in the woods without our smartphones. [51:13] So something has changed there. [51:16] Thank you. [51:17] And I, [51:18] Uh, [51:20] whether we've crossed a sharp boundary or not, I don't know, but I know that, that, that, that I, the frog and beginning to notice the heat. Yep. I feel that, um, you know, I mean, obviously I love playing with new technology and all that kind of stuff, but I feel that too. I, I, um, I,

[51:44] I love books. My favorite thing to do is wake up in the morning, sit on the couch, and just... [51:51] Take a physical book and read it. [51:52] That's how I discovered you. [51:54] Um, and, uh, I'm in my thirties now. And so it's like the first time where I'm starting to see, like, there's a new generation of [52:02] kids and they're just like they don't read [52:04] They don't read like I do, I don't think. Some of them do, but most of them don't. And you see them scrolling on TikTok or whatever. And I've had that first feeling of like, oh, shit. I'm getting a little older and they're doing stuff that I don't know if it's so good. They should be reading. [52:19] And one of the things that I've been playing with, I don't know if this is right, but one of the things I've been playing with is... [52:24] It is true and it is uncomfortable for me, but I know, for example, that [52:29] my brain [52:31] is [52:32] different because I grew up with books. It's different from someone that didn't. And I'm [52:38] their brain, even though they're not totally recognizable to me, it has adapted in this particular way. Like we haven't, we haven't, [52:46] figured out what the limits of human brains are because we've never [52:50] tested them. [52:51] as far as they can go because we haven't had the techno like [52:55] what you grow up with determines the level of, [52:58] you know, technological facility or intelligence or whatever, whatever, um, that you have. And that's why IQ scores go up. [53:06] pretty linearly over time. [53:09] Uh, so, uh, [53:11] We could be the frog, but also each generation of frogs is like slightly better at adapting to temperature than the generation before. So even though the generation before is...

[53:23] is a little bit hot. The new generation is maybe... [53:28] Okay, is my more optimistic take maybe, or my thought. [53:33] Yeah, but at some point, all those frogs, even the ones that are more adapted, will [53:39] the higher temperatures, they're all going to be dead. [53:46] I can't argue with that. I can't argue with that unless the frogs invent, you know, technology that submarines, you know, this is, this has been a fantastic conversation. Is there anything else that you wanted to talk about that, that we didn't get to today? [54:04] Well, we didn't say very much about [54:07] The new book, [54:08] the miraculous from the material. And I would just say a few words about it. That it's, it's, it's got about 35 chapters and each chapter begins with a full page, [54:19] color photograph of an extraordinary visual phenomena like [54:23] spider web or lightning or [54:27] So bubbles. [54:29] And and then there's an essay accompanying the photograph that. [54:35] not only explains the science behind it, but explains my personal experience with that thing. [54:41] and so that's [54:45] That's a brief description of the book. I think it's a wonderful book. If you're listening to this, you should go pick it up. And yeah, it was great to get to chat with you. Thank you so much for coming on.

[54:56] Well, I'm very happy to be on your program, Dan Shipper, and you are a real thinker yourself. You're not an amoeba. [55:06] Thank you. [55:26] It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure unadulterated knowledge bombs about chat GPT. Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat. [55:41] craving for more. It's not just a show. It's a journey into the future with Dan Shipper as the captain of the spaceship. So do yourself a favor. Hit like, smash subscribe and strap in for the ride of your life. [55:55] And now, without any further ado, let me just say, Dan, I'm absolutely hopelessly in love with you.